Datasets:
GEM
/
SciDuet

Tasks:
Other
Languages: English
Multilinguality: unknown
Size Categories: unknown
Language Creators: unknown
Annotations Creators: none
Source Datasets: original
License: apache-2.0
Sebastian Gehrmann commited on
Commit
e36120a
1 Parent(s): bf8f4d2
Files changed (2) hide show
  1. SciDuet.json +7 -4
  2. validation.json +0 -0
SciDuet.json CHANGED
@@ -4,10 +4,10 @@
4
  "has-leaderboard": "no",
5
  "leaderboard-url": "N/A",
6
  "leaderboard-description": "N/A",
7
- "website": "https://huggingface.co/datasets/GEM/SciDuet",
8
- "data-url": "https://github.com/IBM/document2slides/tree/main/SciDuet-ACL",
9
- "paper-url": "https://aclanthology.org/2021.naacl-main.111/",
10
- "paper-bibtext": "@inproceedings{sun-etal-2021-d2s,\n title = \"{D}2{S}: Document-to-Slide Generation Via Query-Based Text Summarization\",\n author = \"Sun, Edward and\n Hou, Yufang and\n Wang, Dakuo and\n Zhang, Yunfeng and\n Wang, Nancy X. R.\",\n booktitle = \"Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies\",\n month = jun,\n year = \"2021\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2021.naacl-main.111\",\n doi = \"10.18653/v1/2021.naacl-main.111\",\n pages = \"1405--1418\",\n abstract = \"Presentations are critical for communication in all areas of our lives, yet the creation of slide decks is often tedious and time-consuming. There has been limited research aiming to automate the document-to-slides generation process and all face a critical challenge: no publicly available dataset for training and benchmarking. In this work, we first contribute a new dataset, SciDuet, consisting of pairs of papers and their corresponding slides decks from recent years{'} NLP and ML conferences (e.g., ACL). Secondly, we present D2S, a novel system that tackles the document-to-slides task with a two-step approach: 1) Use slide titles to retrieve relevant and engaging text, figures, and tables; 2) Summarize the retrieved context into bullet points with long-form question answering. Our evaluation suggests that long-form QA outperforms state-of-the-art summarization baselines on both automated ROUGE metrics and qualitative human evaluation.\",\n}",
11
  "contact-name": "N/A",
12
  "contact-email": "N/A"
13
  },
@@ -35,6 +35,9 @@
35
  "structure-labels": "The original papers and slides (both are in PDF format) are carefully processed by a combination of PDF/Image processing tookits. The text contents from multiple slides that correspond to the same slide title are mreged.",
36
  "structure-splits": "Training, validation and testing data contain 136, 55, and 81 papers from ACL Anthology and their corresponding slides, respectively. ",
37
  "structure-splits-criteria": "The dataset integrated into GEM is the ACL portion of the whole dataset described in \"https://aclanthology.org/2021.naacl-main.111.pdf\", It contains the full Dev and Test sets, and a portion of the Train dataset. \nNote that although we cannot release the whole training dataset due to copyright issues, researchers can still use our released data procurement code to generate the training dataset from the online ICML/NeurIPS anthologies."
 
 
 
38
  }
39
  },
40
  "curation": {
4
  "has-leaderboard": "no",
5
  "leaderboard-url": "N/A",
6
  "leaderboard-description": "N/A",
7
+ "website": "[Huggingface](https://huggingface.co/datasets/GEM/SciDuet)",
8
+ "data-url": "[Github](https://github.com/IBM/document2slides/tree/main/SciDuet-ACL)",
9
+ "paper-url": "[ACL Anthology](https://aclanthology.org/2021.naacl-main.111/)",
10
+ "paper-bibtext": "```\n@inproceedings{sun-etal-2021-d2s,\n title = \"{D}2{S}: Document-to-Slide Generation Via Query-Based Text Summarization\",\n author = \"Sun, Edward and\n Hou, Yufang and\n Wang, Dakuo and\n Zhang, Yunfeng and\n Wang, Nancy X. R.\",\n booktitle = \"Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies\",\n month = jun,\n year = \"2021\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2021.naacl-main.111\",\n doi = \"10.18653/v1/2021.naacl-main.111\",\n pages = \"1405--1418\",\n abstract = \"Presentations are critical for communication in all areas of our lives, yet the creation of slide decks is often tedious and time-consuming. There has been limited research aiming to automate the document-to-slides generation process and all face a critical challenge: no publicly available dataset for training and benchmarking. In this work, we first contribute a new dataset, SciDuet, consisting of pairs of papers and their corresponding slides decks from recent years{'} NLP and ML conferences (e.g., ACL). Secondly, we present D2S, a novel system that tackles the document-to-slides task with a two-step approach: 1) Use slide titles to retrieve relevant and engaging text, figures, and tables; 2) Summarize the retrieved context into bullet points with long-form question answering. Our evaluation suggests that long-form QA outperforms state-of-the-art summarization baselines on both automated ROUGE metrics and qualitative human evaluation.\",\n}\n```",
11
  "contact-name": "N/A",
12
  "contact-email": "N/A"
13
  },
35
  "structure-labels": "The original papers and slides (both are in PDF format) are carefully processed by a combination of PDF/Image processing tookits. The text contents from multiple slides that correspond to the same slide title are mreged.",
36
  "structure-splits": "Training, validation and testing data contain 136, 55, and 81 papers from ACL Anthology and their corresponding slides, respectively. ",
37
  "structure-splits-criteria": "The dataset integrated into GEM is the ACL portion of the whole dataset described in \"https://aclanthology.org/2021.naacl-main.111.pdf\", It contains the full Dev and Test sets, and a portion of the Train dataset. \nNote that although we cannot release the whole training dataset due to copyright issues, researchers can still use our released data procurement code to generate the training dataset from the online ICML/NeurIPS anthologies."
38
+ },
39
+ "what": {
40
+ "dataset": "This dataset supports the document-to-slide generation task where a model has to generate presentation slide content from the text of a document. "
41
  }
42
  },
43
  "curation": {
validation.json CHANGED
The diff for this file is too large to render. See raw diff